Student mathematics, computer science: Consistent Multi-view Texturing of Large-scale 3D City Models

Your mission:

In classical dense stereo matching of aerial imagery, the result of two or more 2D input images is a dense point cloud in 3D space. For various applications, one often needs to mesh the point cloud to a triangulated closed surface, which then can be textured to produce visually appealing results. Well known examples are the 3D views of e.g. Google Maps or Bing Maps. However, assigning a texture to a complex 3D scene, based on multiple camera views, poses some difficulties:

  • Full or partial occlusions of single triangles in the camera views
  • Non-perpendicular looking angles lead to distorted textures
  • Different illumination conditions in the camera views yield patchwork results, exhibiting strong color discontinuities along patch borders (see Figure 1 and 2)

The goal of this thesis is to devise a global energy based framework (e.g. using graph cuts), creating a combined texture map which minimizes the bovementioned errors. Implementation is done using and extending an existing C++ framework. Special care needs to be taken when addressing performance issues, as the data amount is large.

Tasks:

  • Get an overview of state-of-the-art algorithms and choose well-suited algorithms
  • Devise quality measures for the texture extraction and embed into global energy
  • minimization framework
  • Create combined texture map based on triangle packing of the extracted textures
  • Color correction along patch boundaries as post-processing step

Your qualifications:

Wird per js gefüllt...

Your benefits:

Look forward to a fulfilling job with an employer who appreciates your commitment and supports your personal and professional development. Our unique infrastructure offers you a working environment in which you have unparalled scope to develop your creative ideas and accomplish your professional objectives. Disabled applicants with equivalent qualifications will be given preferential treatment.

Course paper / final thesis

immediate

6 months

DLR Oberpfaffenhofen

To location

DLR Remote Sensing Technology Institute

To institute

Contact person :

Georg Kuschk

Earth Observation Center Tel: +49 8153 28-2652

Send message

Course paper / final thesis

Career opportunities

Exciting challenges for inquisitive people: write your thesis at DLR!

Exciting challenges for inquisitive people: write your thesis at DLR! Copyright: Photo.com, Jupiterimages

Two ways to apply

Application

Your way to DLR: here's how you should apply.

SpaceJustin – your soon-to-be new colleague?